Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

BuresTomFlow: Bures-Geometric Flow Matching for Posterior Quantum Tomography

Lu Wei, Yufeng Wang, Chenfeng Cao, Haibin Ling

Published Nov 11, 2026• 0 citations
Publisher

Abstract

Finite-shot quantum tomography often leaves a posterior over many density matrices compatible with the same measurement record, but fast reconstruction methods collapse this uncertainty to a single state. We frame this setting as amortized posterior sampling and introduce BuresTomFlow, a measurement-conditioned Flow Matching sampler for full-rank density matrices. The method transports base density matrices along Bures-Wasserstein paths on the positive matrix cone, projects the path back to the trace-one state space, and trains the velocity field with a Bures-aligned tangent loss. Thus the sampler is trained in the same fidelity-based geometry used to judge posterior quality. We evaluate BuresTomFlow in dense four-qubit simulations with count, shadow, and mixed measurement records. Against Cholesky, log-density, and generic Riemannian Flow Matching controls at comparable learned-sampler runtime, BuresTomFlow improves Bures distance, credible-ball coverage, and held-out observable prediction. In the main comparison across multiple seeds, mean Bures distance decreases from 0.618 for the Cholesky baseline to 0.367 for BuresTomFlow, and mean coverage error decreases from 0.203 to 0.100. These results show that the value of Bures geometry is not merely enforcing physical states, but producing calibrated posterior uncertainty when finite-record tomography cannot be reduced to one reconstruction.

DOI
10.20944/preprints202611.1144.v1

Related papers

GRPO-QPS: Target-Preserving Reinforcement Learning for Quantum Posterior Sampling

Yufeng Wang, Parivesh Priye, Lu Wei +1 more

Bayesian quantum tomography requires efficient inference while preserving a posterior fixed by the prior and Born likelihood. Learned transport provides fast amortized samples, but reward tuning can reshape the generated distribution rather than improve exploration of this fixed target. We introduce GRPO-QPS, a target-preserving framework in which GRPO learns proposal behavior and an exact Metropolis correction preserves the posterior after training. Across the evaluated reconstruction benchmarks, GRPO-QPS improves over BuresTomFlow and Flow-GRPO on thermal, cat, Dicke, and cluster families, and it closely matches an exact two-qubit reference posterior. Tuned conventional MCMC is slightly stronger on several original continuous benchmarks where the available fixed proposals already match the posterior geometry well. To test whether this reflects a fundamental limitation of learned exploration, we evaluate a more challenging multimodal thermal posterior. At six qubits and 800 shots, the learned proposal achieves a minimum effective sample size of 102 per $1{,}000$ likelihood calls, compared with 28 for prior independence, 27 for a tuned fixed mixture, and 20 for Haario adaptive Metropolis. A record-conditioned policy also transfers to unseen 3,000-shot records, matching or exceeding the strongest conventional baseline in all nine held-out seed-record comparisons. These results show that GRPO-QPS combines target-preserving Bayesian inference with broad gains over learned transport baselines and a sampling advantage when efficient exploration requires proposal geometry beyond the evaluated conventional kernels.

cs.LGCross-listedarXiv:2609.14711•Sep 15, 2026•0 citations

Machine learning models for quantum state tomography

Marko Čechovič

This thesis asks whether modern machine-learning model families and restoration ideas that have driven progress in computer graphics can also improve quantum state tomography, the reconstruction of an unknown quantum state from repeated measure- ments. We study four broad families-feed-forward neural networks, conditional genera- tive adversarial networks, iterative generative models from diffusion to flow matching, and transformers-across three practical regimes: incomplete near-clean measurements, com- plete finite-shot measurements, and measurements that are both incomplete and noisy. Five selected point-estimation pipelines are developed through validation-guided exper- iments and evaluated on the same synthetic two-qubit state partitions and paired test records. The common formulation treats missing settings as masked data, represents shot-dependent reliability through confidence, and uses a normalised Cholesky map to ensure a physical output. Combined with set-based aggregation and Bures-based train- ing, this formulation reduces the complete high-shot mean Bures distance of a naive multilayer perceptron from 0.079 to 0.022. On this common in-distribution benchmark, RhoTransformer has the lowest mean Bures distance at all four headline operating points, reaching 0.015 with all sixteen settings at 105...

•Jan 1, 2026•0 citations

Geometric Latent Space Tomography with Metric-Preserving Autoencoders

S. M. Yousuf Iqbal Tomal, Abdullah Al Shafin

Quantum state tomography faces exponential scaling with system size, while recent neural network approaches achieve polynomial scaling at the cost of losing the geometric structure of quantum state space. We introduce geometric latent space tomography, combining classical neural encoders with parameterized quantum circuit decoders trained via a metric-preservation loss that enforces proportionality between latent Euclidean distances and quantum Bures geodesics. On two-qubit mixed states with purity 0.85--0.95 representing NISQ-era decoherence, we achieve high-fidelity reconstruction (mean fidelity $F = 0.942 \pm 0.03$) with an interpretable 20-dimensional latent structure. Critically, latent geodesics exhibit strong linear correlation with Bures distances (Pearson $r = 0.88$, $R^2 = 0.78$), preserving 78\% of quantum metric structure. Geometric analysis reveals intrinsic manifold dimension 6.35 versus 20 ambient dimensions and measurable local curvature ($κ= 0.011 \pm 0.006$), confirming non-trivial Riemannian geometry with $O(d^2)$ computational advantage over $O(4^n)$ density matrix operations. Unlike prior neural tomography, our geometry-aware latent space enables direct state discrimination, fidelity estimation from Euclidean distances, and interpretable error manifolds for quantum error mitigation without repeated full tomography, providing critical capabilities for NISQ devices with limited coherence times.

quant-pharXiv:2512.15801•Dec 18, 2025•0 citations

A Langevin sampler for quantum tomography

Tameem Adel, Abhishek Agarwal, Stéphane Chrétien +4 more

Quantum tomography involves obtaining a full classical description of a prepared quantum state from experimental results. We propose a Langevin sampler for quantum tomography, that relies on a new formulation of Bayesian quantum tomography exploiting the Burer-Monteiro factorization of Hermitian positive-semidefinite matrices. If the rank of the target density matrix is known, this formulation allows us to define a posterior distribution that is only supported on matrices whose rank is upper-bounded by the rank of the target density matrix. Conversely, if the target rank is unknown, any upper bound on the rank can be used by our algorithm, and the rank of the resulting posterior mean estimator is further reduced by the use of a low-rank promoting prior density. This prior density is a complex extension of the one proposed in (Annales de l'Institut Henri Poincare Probability and Statistics, 56(2):1465-1483, 2020). We derive a PAC-Bayesian bound on our proposed estimator that matches the best bounds available in the literature, and we show numerically that it leads to strong scalability improvements compared to existing techniques when the rank of the density matrix is known to be small.

arXiv:2601.08775•Jan 13, 2026•0 citations

A Design Space Study of Density Matrix Parameterizations for Diffusion-Based Quantum State Tomography

Shuangu Chang

Diffusion-based quantum state tomography (QST) has shown promising results, but all existing methods implicitly adopt a single parameterization (typically Cholesky) without systematic evaluation. We present the first design space study of density matrix parameterizations for diffusion QST, introducing a geometric framework based on the Jacobian Gram matrix $\mathbf{J}^\top\mathbf{J}$. Our calibration of seven parameterizations at 2- and 3-qubit scales, validated by end-to-end training, reveals that \emph{geometric conditioning alone does not predict end-to-end performance}: at 3-qubit scale, Hermitian direct ($κ= 2.0\times$) performs worse than Cholesky ($κ= 27\times$) at all shot levels---a $13.5\times$ isotropy advantage that translates into a fidelity \emph{disadvantage} of up to $+0.51$. The 2-qubit ranking (Hermitian $>$ Bloch) reverses at 3 qubits (Bloch 0.907 vs.\ Hermitian 0.394). We provide a geometric explanation: unbounded parameterizations suffer projection-induced information loss because the PSD constraint couples diagonal and off-diagonal coordinates in ways the unconstrained model cannot respect, whereas the Bloch representation places the maximally mixed state at the center of the valid region, minimizing projection loss.

quant-pharXiv:2608.09625•Aug 11, 2026•0 citations

Ricci flow for the Bures--Helstrom qubit metric

Andrew Lesniewski

The Bures--Helstrom metric is the minimal monotone Riemannian metric on the state space of a qubit. With the quantum Fisher normalization used here, it identifies the Bloch ball with a geodesic hemisphere of the unit round three--sphere. We describe its Ricci flow explicitly. In a general rotationally symmetric gauge the flow is a coupled system for the radial lapse and warping factor; a single scalar equation appears only after a Hamilton--DeTurck gauge choice. In the corresponding moving DeTurck frame the squared warping function $Ψ=Φ^2$ satisfies the linear forced heat equation \begin{equation*} D_tΨ=Ψ_{ss}-2, \end{equation*} while the fixed-lapse coordinate form contains the associated transport term. Since the Bures--Helstrom metric is Einstein, the geometric flow itself is the homothetic shrinker \begin{equation*} g(t)=(1-4t)g_{\mathrm{BH}}, \end{equation*} with scalar curvature $6/(1-4t)$ and extinction time $T=1/4$. Thus the metric remains inside the monotone cone for all $t<T$ and leaves the cone of nondegenerate Riemannian metrics only through the collapsed limit. We also record the volume--normalized flow, for which the Bures--Helstrom metric is a fixed point. Its linearization is the shifted round--sphere Laplacian $Δ_{\mathbb S^3}+3$, with spectrum \begin{equation*} σ_\ell=-(\ell-1)(\ell+3), \end{equation*} and spectral gap $5$ after removal of the scaling mode.

cs.ITCross-listedarXiv:2606.19493•Jun 19, 2026
Finite-shot quantum tomography often leaves a posterior over many density matrices compatible with the same measurement record, but fast reconstruction methods collapse this uncertainty to a single state. We frame this setting as amortized posterior sampling and introduce BuresTomFlow, a measurement-conditioned Flow Matching sampler for full-rank density matrices. The method transports base density matrices along Bures-Wasserstein paths on the positive matrix cone, projects the path back to the trace-one state space, and trains the velocity field with a Bures-aligned tangent loss. Thus the sampler is trained in the same fidelity-based geometry used to judge posterior quality. We evaluate BuresTomFlow in dense four-qubit simulations with count, shadow, and mixed measurement records. Against Cholesky, log-density, and generic Riemannian Flow Matching controls at comparable learned-sampler runtime, BuresTomFlow improves Bures distance, credible-ball coverage, and held-out observable prediction. In the main comparison across multiple seeds, mean Bures distance decreases from 0.618 for the Cholesky baseline to 0.367 for BuresTomFlow, and mean coverage error decreases from 0.203 to 0.100. These results show that the value of Bures geometry is not merely enforcing physical states, but producing calibrated posterior uncertainty when finite-record tomography cannot be reduced to one reconstruction.
Bayesian quantum tomography requires efficient inference while preserving a posterior fixed by the prior and Born likelihood. Learned transport provides fast amortized samples, but reward tuning can reshape the generated distribution rather than improve exploration of this fixed target. We introduce GRPO-QPS, a target-preserving framework in which GRPO learns proposal behavior and an exact Metropolis correction preserves the posterior after training. Across the evaluated reconstruction benchmarks, GRPO-QPS improves over BuresTomFlow and Flow-GRPO on thermal, cat, Dicke, and cluster families, and it closely matches an exact two-qubit reference posterior. Tuned conventional MCMC is slightly stronger on several original continuous benchmarks where the available fixed proposals already match the posterior geometry well. To test whether this reflects a fundamental limitation of learned exploration, we evaluate a more challenging multimodal thermal posterior. At six qubits and 800 shots, the learned proposal achieves a minimum effective sample size of 102 per 1,0001{,}0001,000 likelihood calls, compared with 28 for prior independence, 27 for a tuned fixed mixture, and 20 for Haario adaptive Metropolis. A record-conditioned policy also transfers to unseen 3,000-shot records, matching or exceeding the strongest conventional baseline in all nine held-out seed-record comparisons. These results show that GRPO-QPS combines target-preserving Bayesian inference with broad gains over learned transport baselines and a sampling advantage when efficient exploration requires proposal geometry beyond the evaluated conventional kernels.
This thesis asks whether modern machine-learning model families and restoration ideas that have driven progress in computer graphics can also improve quantum state tomography, the reconstruction of an unknown quantum state from repeated measure- ments. We study four broad families-feed-forward neural networks, conditional genera- tive adversarial networks, iterative generative models from diffusion to flow matching, and transformers-across three practical regimes: incomplete near-clean measurements, com- plete finite-shot measurements, and measurements that are both incomplete and noisy. Five selected point-estimation pipelines are developed through validation-guided exper- iments and evaluated on the same synthetic two-qubit state partitions and paired test records. The common formulation treats missing settings as masked data, represents shot-dependent reliability through confidence, and uses a normalised Cholesky map to ensure a physical output. Combined with set-based aggregation and Bures-based train- ing, this formulation reduces the complete high-shot mean Bures distance of a naive multilayer perceptron from 0.079 to 0.022. On this common in-distribution benchmark, RhoTransformer has the lowest mean Bures distance at all four headline operating points, reaching 0.015 with all sixteen settings at 105...
Quantum state tomography faces exponential scaling with system size, while recent neural network approaches achieve polynomial scaling at the cost of losing the geometric structure of quantum state space. We introduce geometric latent space tomography, combining classical neural encoders with parameterized quantum circuit decoders trained via a metric-preservation loss that enforces proportionality between latent Euclidean distances and quantum Bures geodesics. On two-qubit mixed states with purity 0.85--0.95 representing NISQ-era decoherence, we achieve high-fidelity reconstruction (mean fidelity F=0.942±0.03F = 0.942 \pm 0.03F=0.942±0.03) with an interpretable 20-dimensional latent structure. Critically, latent geodesics exhibit strong linear correlation with Bures distances (Pearson r=0.88r = 0.88r=0.88, R2=0.78R^2 = 0.78R2=0.78), preserving 78\% of quantum metric structure. Geometric analysis reveals intrinsic manifold dimension 6.35 versus 20 ambient dimensions and measurable local curvature (κ=0.011±0.006κ= 0.011 \pm 0.006κ=0.011±0.006), confirming non-trivial Riemannian geometry with O(d2)O(d^2)O(d2) computational advantage over O(4n)O(4^n)O(4n) density matrix operations. Unlike prior neural tomography, our geometry-aware latent space enables direct state discrimination, fidelity estimation from Euclidean distances, and interpretable error manifolds for quantum error mitigation without repeated full tomography, providing critical capabilities for NISQ devices with limited coherence times.
Quantum tomography involves obtaining a full classical description of a prepared quantum state from experimental results. We propose a Langevin sampler for quantum tomography, that relies on a new formulation of Bayesian quantum tomography exploiting the Burer-Monteiro factorization of Hermitian positive-semidefinite matrices. If the rank of the target density matrix is known, this formulation allows us to define a posterior distribution that is only supported on matrices whose rank is upper-bounded by the rank of the target density matrix. Conversely, if the target rank is unknown, any upper bound on the rank can be used by our algorithm, and the rank of the resulting posterior mean estimator is further reduced by the use of a low-rank promoting prior density. This prior density is a complex extension of the one proposed in (Annales de l'Institut Henri Poincare Probability and Statistics, 56(2):1465-1483, 2020). We derive a PAC-Bayesian bound on our proposed estimator that matches the best bounds available in the literature, and we show numerically that it leads to strong scalability improvements compared to existing techniques when the rank of the density matrix is known to be small.
Diffusion-based quantum state tomography (QST) has shown promising results, but all existing methods implicitly adopt a single parameterization (typically Cholesky) without systematic evaluation. We present the first design space study of density matrix parameterizations for diffusion QST, introducing a geometric framework based on the Jacobian Gram matrix J⊤J\mathbf{J}^\top\mathbf{J}J⊤J. Our calibration of seven parameterizations at 2- and 3-qubit scales, validated by end-to-end training, reveals that \emph{geometric conditioning alone does not predict end-to-end performance}: at 3-qubit scale, Hermitian direct (κ=2.0×κ= 2.0\timesκ=2.0×) performs worse than Cholesky (κ=27×κ= 27\timesκ=27×) at all shot levels---a 13.5×13.5\times13.5× isotropy advantage that translates into a fidelity \emph{disadvantage} of up to +0.51+0.51+0.51. The 2-qubit ranking (Hermitian >>> Bloch) reverses at 3 qubits (Bloch 0.907 vs.\ Hermitian 0.394). We provide a geometric explanation: unbounded parameterizations suffer projection-induced information loss because the PSD constraint couples diagonal and off-diagonal coordinates in ways the unconstrained model cannot respect, whereas the Bloch representation places the maximally mixed state at the center of the valid region, minimizing projection loss.
The Bures--Helstrom metric is the minimal monotone Riemannian metric on the state space of a qubit. With the quantum Fisher normalization used here, it identifies the Bloch ball with a geodesic hemisphere of the unit round three--sphere. We describe its Ricci flow explicitly. In a general rotationally symmetric gauge the flow is a coupled system for the radial lapse and warping factor; a single scalar equation appears only after a Hamilton--DeTurck gauge choice. In the corresponding moving DeTurck frame the squared warping function Ψ=Φ2Ψ=Φ^2Ψ=Φ2 satisfies the linear forced heat equation \begin{equation*} D_tΨ=Ψ_{ss}-2, \end{equation*} while the fixed-lapse coordinate form contains the associated transport term. Since the Bures--Helstrom metric is Einstein, the geometric flow itself is the homothetic shrinker \begin{equation*} g(t)=(1-4t)g_{\mathrm{BH}}, \end{equation*} with scalar curvature 6/(1−4t)6/(1-4t)6/(1−4t) and extinction time T=1/4T=1/4T=1/4. Thus the metric remains inside the monotone cone for all t<Tt<Tt<T and leaves the cone of nondegenerate Riemannian metrics only through the collapsed limit. We also record the volume--normalized flow, for which the Bures--Helstrom metric is a fixed point. Its linearization is the shifted round--sphere Laplacian ΔS3+3Δ_{\mathbb S^3}+3ΔS3​+3, with spectrum \begin{equation*} σ_\ell=-(\ell-1)(\ell+3), \end{equation*} and spectral gap 555 after removal of the scaling mode.